One of the key challenges to exploiting machine learning is the generation of large scale labelled data which can be used to train deep learning models. In many environment such as agriculture there is limited available data and the resource required to capture and label such data is both labour and time intensive. To address this challenge the Loughborough University research team developed a Orchard simulation using Unreal Engine 4 from which training data can be quickly captured and labelled using the UnrealCV plugin. This method of generating training provided increases in the accuracy of the computer vision system employed in the project and provides a basis for future research in this area.